学长，关于单变量截断分布的抽样你可以参考下面的方法：
Algorithm 5.1: Probability integral transform method
5.1 Draw u from U(0, 1).
5.2 Return y = F−1(u) as a draw from f (y).
An important application of this method is to the problem of sampling from a
truncated distribution. Suppose that X has d.f. F(X) and that we wish to generate
values of X restricted to c1 ≤ X ≤ c2. The distribution of the truncated values is
[F(X) − F(c1)]/[F(c2) − F(c1)] for c1 ≤ X ≤ c2. Following our usual procedure,
we generate U ∼ U(0, 1) and set
U = (F(X) − F(c1))/(F(c2) − F(c1)),which implies that
X = F^-1(F(c1) + U[F(c2) − F(c1)])   (最后这个是反函数，因为懒得用公式编辑器了，呵呵)

您好。若是多元截断正态分布，怎么生成呢？比如生成在u<-rep(1,N)处截断的服从N(u,epsion)的随机数，其中epsion为协方差矩阵？

您好。若是多元截断正态分布，怎么生成呢？比如生成在u<-rep(1,N)处截断的服从N(u,epsion)的随机数，其中epsion为协方差矩阵？

我只想到了笨办法，把不符合条件的去掉
R有生成截断分布的包，去CRAN查一下吧

这个分布的逆函数如何求呢